Patients with myeloid leukemia relapsing after BMT have been successfully treated with donor lymphocyte transfusions (DLTs). We have established a murine BMT model system in which donor splenocytes infused later post-BMT mediate a potent GVL effect against AML. In aim 1, we will first focus on determining what factors preclude GVHD lethality in this setting. Our preliminary data suggest that an important factor controlling GVHD lethality is the presence of CD4+ T cells at the time of DLTs. Studies will be performed to determine the origin of these CD4+ T cells and will assess how the CD4+ T cells affect the engraftment and function of donor splenocytes infused on day 21 post-BMT. Myelosuppression is an important complication of DLTs. We hypothesize that factors which may increase this risk include the presence of: (1) normal host hematopoietic cells; (2) residual leukemia cells; and (3) a reduced donor hematopoietic cell mass. The relative importance of these factors will be addressed using systems in which nonmalignant or malignant type host hematopoietic cells or a reduced donor hematopoietic cell mass are present at the time of DLTs. In aim 2, we will determine why are DLTs only partially effective in AML and explore strategies to enhance GVL. We will first determine if GVHD is beneficial in providing a donor anti-host (and hence anti-tumor) response or detrimental via immune suppression. Experiments will be performed under conditions in which donor anti-host responses are augmented or precluded. We will also determine whether memory T cell responses to AML cells developed. To determine if AML persistence is beneficial for sustaining memory responses to AML, the level of residual AML will be assessed by semi-quantitative PCR at periodic intervals and residual disease levels will be correlated with memory responses. Situations will be established in which donor T cells incapable of mediating GVHD are exposed to AML cells prior to or after delayed infusion. To extend memory cell responses and T cell longevity, donor T cells will be transduced with genes which constituitively signal T cells, retard apoptotic cell death, or provide an autocrine growth factor prior to infusion. Because these vectors contain cell surface expressed determinants, transduced T cells can be positively selected prior to infusion and retroviral expression can be monitored after in vivo infusion.